Road-mapping the Way Forward for Sentinel-3 Topography Mission SAR-Mode waveform Re-tracking over water surfaces J Benveniste, P Cipollini, C Gommenginger, C Martin-Puig, D Cotton, S Dinardo, NOC (UK), Starlab (Spain), SatOC (UK), ESA/ESRIN (Italy)

Introduction This talk deals with findings and challenges relevant to the Sentinel-3 Radar Altimeter over ocean We will see: What is special about the Sentinel-3 Altimeter What are the challenges Validation Exercise: height inter-calibration with Jason-2 and SWH calibration w.r.t. wave buoys Conclusions

Conventional Low Resolution Mode (LRM) Vs/c PRF: ~2 kHz In LRM, the pulse-limited footprint moves across the surface pulse by pulse as the satellite passes overhead The footprint of a conventional pulse-limited altimeter moves along the surface at (essentially) the same rate as the progress of the satellite along its orbit. Typically, the altimeter transmits a pulse 1000 to 2000 times per second (the so-called pulse repetition frequency). Since the speed of the satellite (projected onto the surface) is about 6.7 km/s, the effective along-track length of the footprint is increased in proportion to the number of pulses so averaged. The half-power width of the antenna beamwidth projected onto the surface is (typically) about 15 km, which is several times larger than the pulse-limited footprint. 3-7 km depending on sea state Courtesy: K. Raney, APL/JHU

SAR altimeter (Cryosat-2, Sentinel-3): fundamentally different! Vs/c PRF: ~20 kHz SAR altimeter uses Doppler frequency effect to spotlight a small area along-track (250-300 m) resolved by each burst as the satellite passes overhead A delay-Doppler radar altimeter uses signal processing techniques to resolve a small along-track footprint. The altimeter in effect “stares” at each resolved surface location for as long as that particular cell is illuminated by the antenna as the spacecraft passes over head. The minimum size of such a delay-Doppler-resolved cell is about 250 meters, which is a constant (determined by system parameters and viewing geometry). Note that each cell is viewed over a larger fraction of the antenna beam than the pulse-limited; thus more data is gathered, which leads to substantial benefits (reviewed in subsequent materials). If only one such cell were illuminated, the measurements could not keep up with the forward velocity of the antenna footprint. The system gets around this objection by operating many of these spotlight beams simultaneously. A typical design would generate 64 parallel Doppler beams, but not all of these fall within the main lobe of the antenna. Thus, in practice there will be date from ~40 beams combined in the processor to generate the final (averaged) waveform from each resolved cell. Courtesy: K. Raney, APL/JHU 250-300 meters

At t=0, the SAR footprint is no longer circular but rectangular !! At t=0, the Across Track Resolution is the same as Pulse-Limited Altimetry (Pulse-Limited Diameter) but the footprint size reduces with the time delay RED LRM GREEN SAR t<0 t=0 t>0 The footprint area is no longer constant with time delay

Multi-looking Doppler-beam selection/weighting & incoherent integration of beams from multiple bursts starting at the same ground cell (to reduce the speckle noise) i.e. multilooking 128 gates 1D multi-looked 20Hz waveform Note the peaky waveform shape and the long decaying tail due to varying footprint area in SAR Mode

Predicted advantages of SAR over LRM More independent looks  improved retrieval precision Two-fold improvement according to numerical studies by Jensen & Raney (1998) Finer spatial resolution along track ~300 meters along-track Higher SNR ~10 db more Better performance close to land especially for track ~90° to coastline Less sensitivity to sea state ..to investigate these and other things, ESA has setup…SAMOSA!! (next slide)

THE FUTURE: Sentinel-3 Surface Topography Mission THE PRESENT: Cryosat-2 LRM: most of open ocean + most of land SAR: some oceanic & coastal regions + sea ice SARIN: few pilot regions (ocean & land) + land ice THE FUTURE: Sentinel-3 Surface Topography Mission LRM: open ocean SAR: global coastal ocean + sea ice SAR open loop: ice sheet margins SAR open/closed loop: land & inland waters

The ESA SAMOSA Project: Study funded by ESA STSE, led by David Cotton (SatOC) Starlab, NOC, De Montfort University, DTU Space & expert support from Keith Raney (JHU/APL) Objectives & Methodology Quantify range retrieval accuracy in pulse-limited and SAR mode as a function of significant wave height Develop physically-based models for SAR altimeter ocean waveforms Apply physically-based models to SAR ocean waveforms Done for both simulated and real Cryosat SAR waveforms over ocean Investigate method to reduce SAR mode data to pseudo-LRM Applications to airborne SAR altimeter data, SAR waveforms over inland water (DMU), coastal regions, sea floor mapping (DTU Space),… Development and Validation of SAMOSA SAR re-tracker with Cryosat-2 SAR mode data over ocean

Next-generation SAMOSA models SAMOSA1 gives acceptable fit to Cryosat-2 SAR waveforms BUT… Circular antenna beam, no across-track mispointing Non-physical behaviour for large mispointing angles Over-estimates significant wave height (Hs) Next generation model: SAMOSA2 entirely new physically-based formulation developed by Starlab Accounts for asymmetric antenna beam, ellipticity of the Earth, along- AND across-track mispointing, non-linear ocean wave effects Physically correct response to mispointing Improved fit to Cryosat-2 SAR waveforms BUT… not fully-analytical & computational expensive

SAMOSA3 model Simplification of SAMOSA2 while keeping its advanced features fully-analytical, robust and computationally fast !!

Comparing SAM1, SAM2 and SAM3 With SYMMETRIC antenna beam and no Earth ellipticity effects: SAM3 and SAM1 are equivalent !! These two entirely independent theoretical models converge !

SAMOSA 3 Project: Study funded by ESA STSE, led by David Cotton (SatOC) Starlab, NOC, Objectives & Methodology Develop and characterize SAM3 model (Trade-off TN) Deliver to the Sentinel-3 STM PAD a Detailed Processing Model (DPM) of a SAR ocean waveform re-tracker based on the best SAMOSA model to operationally retrack Sentinel-3 SAR-Mode L1b waveforms Validate Retracker using Cryosat Data in high and low sea conditions

SAMOSA 3 DDM and Re-tracker: Physically-based model developed by Starlab from first principles Analytical ( by Bessel Functions) solutions to model the Delay Doppler Maps (DDM) for the full span of Doppler Frequencies Model depends on epoch, significant wave height, Pu, surface rms slope, and mispointing angle(s), The model independent variables are the Doppler Frequency and the Time Delay The waveforms are retracked by Least- Square Fitting Algorithm (Levenberg-Marquard)

Model Behaviour: effect of Sea State (SWH) SWH affects the width of lobe of the SAR waveforms SAMOSA3 model: Dependence of waveform shape on SWH

Model Behaviour: effect of mispointing Roll mispointing affect shape of the SAR waveforms Model insensible to pitch mispointing Effect significant only for mispointing greater than 0.1 deg

Effect of along-track mispointing: negligible

Effect of across-track (roll) mispointing: changes waveform tail 0.5° 0.3° 0.1° 0.0°

Challenges of SAR altimetry

Challenge 1: LRMSAR transitions

LRMSAR transitions (cont.) Methodology to reduce SAR mode data to pseudo LRM has been developed by Starlab (RDSAR method) Tested and validated with simulated SAR waveforms: Full Bit Rate (FBR) SAR can be successfully transformed into pseudo-LRM waveforms that can be re-tracked with Brown model RDSAR anyway does not give the same range noise as LRM (less number of accumulated independent looks) RDSAR ensures continuity

Challenge 2: Coastal Zones Waveforms at 20 Hz, one waveform each 300 meters – Optimal Conditions Coastal Zone

Effect of the Land -> LIMITS Track Orientation and bright targets can have effect on the level of the Land Contamination.. Even in SAR mode !!! The land contamination in SAR Mode is greatly reduced but still present in unfavourable cases; MisFit and Ground Track Orientation to the coastline can be parameters useful to quantify the contamination

Challenge 3: Inland Water

Validation Exercise over open sea

CryoSat-2 and Jason-2: Range error in SAR and LRM as a function of sea state Lots of CryoSat-2 data in LRM and SAR mode since July 2010 but not collocated (SAR and LRM are exclusive modes) Comparison of CryoSat2 SAR with Jason2 LRM Focus on small area in Norwegian Sea July 2010 – March 2011 Wide range of sea states

Variability in SSH (Norwegian Sea) SAM3 SSH Ksiyinput = 0.01 deg CryoSat-2 SAR (Hs < 2m) SSH std 20Hz: 4.907 cm 1Hz: 1.097 cm Jason-2 LRM (Hs < 2m) SSH std 20Hz: 7.012 cm 1Hz: 1.568 cm Factor of 1.43 reduction in SSH error with CryoSat-2 L1B SAR SAR perf will improve with reprocessed CryoSat-2 L1B data

Variability in Hs (Norwegian Sea) SAM3 SSH Ksiyinput = 0.01 deg CryoSat-2 SAR (Hs < 2m) Hs std 20Hz: 0.425 m 1Hz: 0.095 m Jason-2 LRM (Hs < 2m) Hs std 20Hz: 0.536 m 1Hz: 0.120 m Slight bias in SAR Hs ? Factor of 1.26 reduction in SWH error with C2 L1B SAR

Bias in SAR Hs ? Validation against buoys Buoy wave height data from UK Met Office CryoSat-2 SAR data and buoys collocated within 50km and 30 minutes Data for July 2010 - May 2011

CryoSat-2 SAR Hs against buoys Linear relation between C2 SAR Hs and buoy Hs over full range of sea states Small bias < 0.5 m; Std ~ 0.3 m Hs bias is directly linked to roll mispointing used as input to the SAR retracker use of platform mispointing data as input to the SAR retracker should reduce this bias Need to account for known small systematic bias in roll mispointing for C2 (Scharroo, pers comm.) SAM3 SWH Ksiyinput = 0.01 deg

Conclusions Performance of CryoSat-2 SAR should improve with optimised SAR retracking that uses platform roll mispointing as input Both variability and SWH bias Same analyses to be repeated in a less dynamic region Analyses of CryoSat-2 SAR performance against pseudo-LRM products In summary: CryoSat-2 SAR mode produces very good data over ocean we have a fully-analytical, robust and computationally efficient model that successfully retracks CryoSat-2 SAR waveforms SAR shows improved performance compared to LRM for both SSH and SWH SAR opens up a wide range of new scientific possibilities over ocean, coastal zone and inland water Ready for Sentinel-3!

www.altimery2012.org